Voices from the Past: AI-Driven Storytelling from the Perspective of Cultural Artifacts

Voices from the Past is an AI-powered application that brings cultural artifacts to life by giving them a voice to tell their own stories. Using AI-generated first-person narratives, users can explore the journeys of historical objects, engage in conversations with them, and discover how facts and imagination blend to create a unique experience.

About the Project

This project was developed as part of a university project for a Digital Humanities course. The goal is to explore the intersection of generative AI and cultural heritage by offering a new, interactive way to engage with history. It aims to make museum pieces more relatable and engaging through storytelling and conversation.

How It Works

• Users can search for artifacts via the Europeana API, which retrieves metadata and images.
• To ensure more predictable and relevant results, the search is currently restricted to three selected museums, offering a diverse range of artifacts.
• A pre-trained language model is used to generate a first-person narrative based on the artifact's metadata.
• Additional factual information is gathered from the artifact‘s webpage, if available.
• Users can chat with the artifact to learn more about its history and cultural significance.
• The application is built with Streamlit to offer an intuitive user interface.

Future Enhancements

Potential improvements include:

• Expanding artifact sources accross and beyond Europeana
• Refining narrative accuracy with improved data sources
• Paying special attention to sensitive arteficts and how to represent them respectfully
• Enhancing user interaction with multimedia elements like audio or video

Hosted App

Try the app on Streamlit Community Cloud: https://pastvoicesai.streamlit.app

You will need to provide access to an OpenAI API compatible model in the Settings.

Local Installation

To run the project locally, you need to provide a model as well as a valid API-key for Europeana.

1. Clone the repository

   git clone https://github.com/katzmo/PastVoicesAI.git
   cd PastVoicesAI

2. Install dependencies Run pip in a virtual python environment, or use another package manager of your choice.

   pip install -r requirements.txt

3. Set up API keys in a new file .streamlit/secrets.toml

   # Europeana API
   EUROPEANA_API_KEY = "your-api-key"
   # OpenAI API
   OPENAI_BASE_URL = "https://api.openai.com/v1"
   OPENAI_API_KEY = "my-openai-key"

   The model API only needs to be compatible with OpenAI, you are not required to use OpenAI!

4. Run the streamlit application

   streamlit run app.py

Disclaimer

This project is an experimental student work and is not intended to provide historically verified information. The AI-generated content is based on available data and creative storytelling, meaning some details may not be historically accurate. The focus is on users engaging with cultural heritage objects in a playful way and having fun testing different AI models, responses should not be taken too seriously.